--- Richard Wordingham
<
richard.wordingham@...> wrote:
> Doesn't rule 8 (see Cybalist message 34520 for a
> definition) leave a
> permanent see-saw between **kjalna`bnt and
> **kjalnba`nt? I'd go for
> vowel epenthesis.
I understand what you mean, but it isn't cyclic or
anything similar. If nasal (or w) metathesis creates
an unacceptable cluster (I haven't worked the
conditions out exactly, but likely there's none if n
moves before dental or any kind of velar (with ng as
allophone of n)) the following vowel moves between n
and the directly following C. If there is no
following vowel, the preceding vowel moves instead.
No other rules come between this. For example, in
"knee" ga'nalmv>ga'nalw>ga'nawl and instead of w>u
there's V-metathesis. Same in "sun".
As for *kjlana`bnt (I really prefer l in the onset, on
further consideration, but don't have much evidence),
at this stage 3rd pl appears as -nt after V and -ant
after C, so an immediate reanalys doesn't seem
unlikely to me.
> The expression of the zero-grading rules is
> confusing, if not
> irredeemably inadequate. I end up with *k^.lnwm�
> (*k^l.nwm� without
> resyllabification) if I start from *k^lab, but then
> 4 sonorants in a
> row is quite a challenge!
Yes. Do you mean the expression in this post? This
is just shorthand for the previously expressed cyclic
a-deletion rules. I'll send the exact section to
clarify.
_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com